Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Google's live from Paris event private/deleted immediately

news.ycombinator.com

Post reply on HN
Print view View original

341–350 of 391 posts

Re: Google's live from Paris event private/deleted immediately

#341

Post by dhdaadhd »

vineyardmike wrote: post #311
sho_hn wrote: post #245

Earlier quoted context omitted.

I'd like to have more discussion about the attribution problem, specifically. We now have a couple of the players saying they're working on, or having demos. But from what I can see, all of these don't _actually_ attribute the source. They're just able to find _a_ source that fits to the output, working backward from it. Often the attributed source fits, but is actually disagreeing with the original output in specifi…

> But from what I can see, all of these don't _actually_ attribute the source. They're just able to find _a_ source that fits to the output, working backward from it I feel like it should be possible to build attribution "simply" by finding good results via traditional search (or even using LLM embeddings), then asking a LLM to summarize the sources. Then you can show the attribution. This would (to me) be much more…

I think Neeva is doing this!

Re: Google's live from Paris event private/deleted immediately

#342

Post by onion2k »

rchaud wrote: post #199
onion2k wrote: post #181

Earlier quoted context omitted.

I don't know that end-users really care about having a GPT-written paragraph accompanying their search results. For 90% of my searches I don't want any search results. I want a definitive, authoritative answer to a question. In the olden days of the web I had to figure out what to search to give me a website that would give me that answer as the first result, and in the mid-90s that was really hard because search was…

> Over the next 20 years the war between Google and SEO spammers meant finding things got a bit harder. Along with that Google's proliferation of adverts on their search result pages meant it actually became harder to even find the website link I wanted in the page itself. Yes, SEO spam has largely made the search experience a lot worse and ripe for change. But ChatGPT crawls these terrible links too! Many of the lin…

So why would you trust GPT to provide an authoritative answer more than your own instincts?

I wouldn't right now, but I believe the accuracy of an LLM is a function of its size and how much you can reduce the loss function. I think OpenAI can improve those things faster than Google based on the things that have been shown so far. I could be completely wrong though.

I also suspect there's no benefit to corrupting a GPT model except for the lulz. If a search tool isn't showing links then there's no financial reason to put effort into setting up link networks and SEO content farms. GPT based search has an innate anti-spam advantage because it renders spam futile if the spam isn't getting visitors.

It also renders a lot of non-spam content futile too, which is problematic. It's not a magic bullet.

Re: Google's live from Paris event private/deleted immediately

#343

Post by CyberDildonics »

amelius wrote: post #205
Tepix wrote: post #179

Earlier quoted context omitted.

It will be integrated with Search, as they have announced.

Yeah, but it will cost them a lot more to run it.

Cannibalize means customers migrate to one thing and leave the other behind.

Re: Google's live from Paris event private/deleted immediately

#344

Post by voisin »

1vuio0pswjnm7 wrote: post #288
PaulDavisThe1st wrote: post #243

You know, whatever the merits (and it sounds as if there were very few) of Google's presentation (or Pinchai's recent non-statement) might be, I would have expected the commentariat here at HN to display way more skepticism about language models as a replacement for search. If there's a crowd anywhere in the world large enough to leave 200+ comments on a post like this, you'd think thatg the one at HN would understan…

Assuming that Google is in a panic over ChatGPT, as suggested the HN commentariat and the "tech" news media, then arguably this only highlights how little confidence Google has in the quality of its search engine and the trustworthiness of the web as a source for information. For example, instead of trying to match or beat ChatGPT, whatever that would entail, Google could instead focus on illustrating the shortcoming…

> arguably this only highlights how little confidence Google has in the quality of its search engine and the trustworthiness of the web as a source for information.

I think they are realizing they killed their golden goose by filling it with trash ads and it only worked in the absence of something more accessible to the public. They were drunk at the wheel and crashed into (for the first time ever) a competent competitor. Now they realize they need to sober up quickly but their organization is lazy and bloated, they’ve taken their users for granted and abused them at the altar of advertisers, and generally done nothing for two entire decades to garner any form of loyalty.

They’ve got some work cut out for them. From my perspective, Google could come out with a ChatGPT killer and I still wouldn’t use it, because the company has a track record of arbitrarily killing off any product I’ve actually liked.

Re: Google's live from Paris event private/deleted immediately

#345

Post by cuttysnark »

ruined wrote: post #132
cuttysnark wrote: post #114

Earlier quoted context omitted.

I had to work with a list of the 50 states abbreviations, full names, etc. Previously, I'd have found the data somewhere with Google et al and then manipulated it myself into the JSON I needed. But this time I asked GPT to not only source the data, but also return it structured to my liking—with revisions ("please include Guam and Puetro Rico") Lots of my searches are simply I want an answer, but often times I also w…

this seems like the kind of thing that gpt would potentially hallucinate something like "East Virginia" or "Ardakota" complete with abbreviation, and maybe that's fine with 50 states but you're playing with fire if it's something you can't immediately catch the mistake also, how hard is it to get a list of abbreviations, and wrap them all in quote and comma and brackets. with vim that's barely more work than a copy-p…

> but you're playing with fire if it's something you can't immediately catch the mistake

Agreed. I was able to manually review 52 results, but wouldn't trust it blindly.

> also, how hard is it to get a list of abbreviations, and wrap them all in quote and comma and brackets.

Not hard at all. And I may still do that, but it was pretty enjoyable to ask the machine.

Re: Google's live from Paris event private/deleted immediately

#346

Post by PaulDavisThe1st »

ekanes wrote: post #322
PaulDavisThe1st wrote: post #243

You know, whatever the merits (and it sounds as if there were very few) of Google's presentation (or Pinchai's recent non-statement) might be, I would have expected the commentariat here at HN to display way more skepticism about language models as a replacement for search. If there's a crowd anywhere in the world large enough to leave 200+ comments on a post like this, you'd think thatg the one at HN would understan…

For me, it's that this recently-released product is already competing (in the minds of users, which is what counts) with a decades-old mature product. In terms of accuracy, the "but sometimes it's wrong" perspective reminds me of the initial reaction among some to wikipedia. Yep, ANYONE can edit it. Yep, they could lie... ¯\_(ツ)_/¯

There are and always have been mechanisms in place to address lying on Wikipedia, which work with varying degrees of effectiveness depending on the topic.

What is the mechanism to address lying by a language model?

Re: Google's live from Paris event private/deleted immediately

#347

Post by cuttysnark »

reaperducer wrote: post #163
cuttysnark wrote: post #114

Earlier quoted context omitted.

I had to work with a list of the 50 states abbreviations, full names, etc. Previously, I'd have found the data somewhere with Google et al and then manipulated it myself into the JSON I needed. But this time I asked GPT to not only source the data, but also return it structured to my liking—with revisions ("please include Guam and Puetro Rico") Lots of my searches are simply I want an answer, but often times I also w…

with revisions ("please include Guam and Puetro Rico") Your wrote "please" to ChatGPT? FWIW, Miss Manners had a column a couple of weeks ago where she stated that we should not say please or thank you to machines, including Siri and Alexa. I was surprised.

I did. I even asked ChatGPT if having "please" and "thank you" included in my prompts affects it in any way shape or form—it told me "No".

I still do it. ¯\_(ツ)_/¯

Re: Google's live from Paris event private/deleted immediately

#349

Post by DecentAI »

Karawebnetwork wrote: post #71
afavour wrote: post #13

ChatGPT as a search engine sounds amazing but also really quite problematic. It's an extension of the issue with Google's "instant answers" (or whatever they're called): right now the creators of the content Google/ChatGPT scrapes are usually paid via the advertisements on their pages. When no-one clicks though any more, no-one gets paid. I know, I know, the ad banner-funded web is a mess and I wouldn't mourn its dem…

Perplexity https://www.perplexity.ai/ does it well in my opinion. It builds a paragraph answering your query but it has a lot of footnotes that link directly to websites. Ex: "geopolitical reason for palm oil being banned and why it's bad for health" The EU has banned palm oil in biofuels due to its negative impacts on health[1] and its geopolitical implications, such as favoring alternative crops grown in Europe[2].…

One of the best AI + search implementations ever

Re: Google's live from Paris event private/deleted immediately

#350

Post by rendall »

PaulDavisThe1st wrote: post #346
ekanes wrote: post #322

Earlier quoted context omitted.

For me, it's that this recently-released product is already competing (in the minds of users, which is what counts) with a decades-old mature product. In terms of accuracy, the "but sometimes it's wrong" perspective reminds me of the initial reaction among some to wikipedia. Yep, ANYONE can edit it. Yep, they could lie... ¯\_(ツ)_/¯

There are and always have been mechanisms in place to address lying on Wikipedia, which work with varying degrees of effectiveness depending on the topic. What is the mechanism to address lying by a language model?

> There are and always have been mechanisms in place to address lying on Wikipedia

Wikipedia's mechanisms are fundamentally unable to cope with motivated, coordinated gangs that conspire together to push a distorted perspective, particularly on niche articles. Given Wikipedia's prominence, this is a problem that will only get worse.

Post reply on HN